1. Field of the Invention
Embodiments described herein relate generally to improved semiconductor imaging devices and in particular to imaging devices having an array of pixels and to methods of operating the pixels to reduce temporal noise.
2. Background of the Invention
A conventional four transistor (4T) circuit for a pixel 150 in a pixel array 230 of a CMOS imager is illustrated in FIG. 1. The 4T pixel 150 has a photosensor such as a photodiode 162, a reset transistor 184, a transfer transistor 190, a source follower transistor 186, and a row select transistor 188. It should be understood that FIG. 1 shows the circuitry for operation of a single pixel 150, and that in practical use, there will be an M×N array of pixels arranged in rows and columns with the pixels of the array 230 being accessed using row and column select circuitry, as described in more detail below.
The photodiode 162 converts incident photons to electrons, which are selectively passed to a floating diffusion region A through the transfer transistor 190 when activated by a TX1 control signal. The source follower transistor 186 has its gate connected to floating diffusion region A and thus amplifies the signal appearing at the floating diffusion region A. When a particular row containing pixel 150 is selected by an activated row select transistor 188, the signal amplified by the source follower transistor 186 is passed on a column line 170 to column readout circuitry (242, FIGS. 2-4). The photodiode 162 accumulates a photo-generated charge in a doped region of its substrate during a charge integration period. It should be understood that the pixel 150 may include a photogate or other photon to charge converting device, in lieu of a photodiode, as the initial accumulator for photo-generated charge.
The gate of transfer transistor 190 is coupled to a transfer control signal line 191 for receiving the TX1 control signal, thereby serving to control the coupling of the photodiode 162 to region A. A voltage source Vpix is selectively coupled through reset transistor 184 and conductive line 163 to floating diffusion region A. The gate of the reset transistor 184 is coupled to a reset control line 183 for receiving a RST control signal to control the reset operation in which the voltage source Vpix is connected to floating diffusion region A.
A row select signal (Row Sel) on a row select control line 160 is used to activate the row select transistor 188. Although not shown, the row select control line 160, reset control line 183, and transfer signal control line 191 are coupled to all of the pixels of the same row of the array. The voltage source Vpix is coupled to transistors 184 and 186 by conductive line 195. The column line 170 is coupled to the output of all of the pixels of the same column of the array and typically has a current sink 176 at one end. Signals from the pixel 150 are selectively coupled to a column readout circuit 242 (FIGS. 2-4) through the column line 170.
As is known in the art, a value can be read from pixel 150 in a two step correlated double sampling process. First, floating diffusion region A is reset by activating the reset transistor 184. The reset signal (e.g., Vrst) found at floating diffusion region A is readout to column line 170 via the source follower transistor 186 and the activated row select transistor 188. During a charge integration period, photodiode 162 produces charge from incident light. This is also known as the image intergration period. After the integration period, the transfer transistor 190 is activated and the charge from the photodiode 162 is passed through the transfer transistor 190 to floating diffusion region A, where the charge is amplified by the source follower transistor 186 and passed to the column line 170 (through the row select transistor 188) as an integrated charge signal Vsig. In some instances, the reset signal Vrst is provided after the integrated charge signal Vsig. As a result, two different voltage signals—the reset signal Vrst and the integrated charge signal Vsig—are readout from the pixel 150 onto the column line 170 and to column readout circuitry 242, where each signal is sampled and held for further processing as is known in the art. Typically, all pixels in a row are readout simultaneously onto respective column lines 170 and the column lines may be activated in sequence or in parallel for pixel reset and signal voltage readout.
FIG. 2 shows an example CMOS imager device 201 that includes the pixel array 230 and a timing and control circuit 232, which provides timing and control signals to enable reading out of signals stored in the pixels in a manner commonly known to those skilled in the art. Example arrays have dimensions of M×N pixels, with the size of the array 230 depending on a particular application. In the illustrated imager device 201, the pixel signals from the array 230 are readout a row at a time using a column parallel readout architecture. The controller 232 selects a particular row of pixels in the array 230 by controlling the operation of row addressing circuit 234 and row drivers 240. Reset Vrst and image Vsig signals in the selected row of pixels are provided on the column lines 170 to a column readout circuit 242 in the manner described above. The signals read from each of the columns can be readout sequentially or in parallel using a column addressing circuit 244. Pixel signals (Vrst, Vsig) corresponding to the readout reset signal and integrated charge signal are provided as respective outputs Vout1, Vout2 of the column readout circuit 242 where they are subtracted in differential amplifier 246, digitized by analog-to-digital converter (ADC) 248, and sent to an image processor circuit 250 for image processing.
FIG. 3 shows more details of one example of the arrangement of the rows and columns 249 of pixels 150 in the array 230. Each column 249 includes multiple rows of pixels 150. Signals from the pixels 150 in a particular column 249 can be readout to sample and hold circuitry 261 associated with the column 249 (part of circuit 242) for acquiring the pixel reset Vrst and integrated charge Vsig signals. Signals stored in the sample and hold circuits 261 can be read sequentially column-by-column to the differential amplifier 246 (FIG. 2), which subtracts the reset and integrated charge signals and sends them to the analog-to-digital converter 248 (FIG. 2). Alternatively, a plurality of analog-to-digital converters 248 may also be provided, each digitizing sampled and held signals from one or more columns 249.
FIG. 4 illustrates portions of three sample and hold circuits 261 of FIG. 3 in greater detail. Each sample and hold circuit 261 holds a set of signals, e.g., a reset signal Vrst and an integrated charge signal Vsig from a desired pixel. For example, a reset signal Vrst of a desired pixel connected to column line 170 is stored on capacitor 226 and the integrated charge signal Vsig from column line 170 is stored on capacitor 228. A front side of capacitor 226 is switchably coupled to the column line 170 through switch 222 and a backside of capacitor 226 is switchably coupled to amplifier 248 through switch 218. A front side of capacitor 228 is switchably coupled to the column line 170 through switch 220 and a backside of capacitor 228 is switchably coupled to amplifier 248 through switch 216. The front side of capacitor 226 is switchably coupled to the front side of capacitor 228 through crowbar switch 239. The backside of capacitor 226 is switchably coupled to the backside of capacitor 228 and to a reference voltage Vref source through clamp switch 299.
Each sample and hold circuit 261 is coupled to amplifier 248 having first and second inputs. The first input of amplifier 248 is coupled to a first output of amplifier 248 through a capacitor 278 and a switch 279 to provide a first feedback circuit. The second input of amplifier 248 is coupled to a second output of amplifier 248 through a capacitor 276 and a switch 277 to provide a second feedback circuit.
The CMOS imager of FIGS. 1-4 has identical correlated double sampling and holding timing for all columns over an entire row. Thus, all of the pixels in a row are readout at substantially the same time. The simplified correlated double sampling and column read out timing is depicted in FIG. 5.
Thus, to begin a readout operation, a logic high clamp signal cl is provided to clamp switch 299 thereby coupling the backsides of capacitors 226, 228 to a reference voltage source Vref. When a reset signal Vrst is read from the pixel 150, a logic high SHR signal is provided to the gate of switch 222 thereby coupling the front side of capacitor 226 to the column line 170. When the readout of the reset signal Vrst from the pixel 150 is complete, a logic low SHR signal is provided to the gate of switch 222 thereby uncoupling the front side of capacitor 226 from the column line 170. Thus, a reset signal Vrst has been sampled and stored on capacitor 226.
After the reset Vrst signal is read from pixel 150, an integrated charge signal Vsig is readout. When the integrated charge signal Vsig is read from pixel 150, a logic high SHS signal is provided to the gate of switch 220 thereby coupling the front side of capacitor 228 to the column line 170. When the readout of the integrated charge signal Vsig from the pixel 150 is complete, a logic low SHS signal is provided to the gate of switch 220 thereby uncoupling the front side of capacitor 228 from the column line 170. Thus, an integrated charge signal Vsig has been sampled and stored on capacitor 228.
When the readout operation is complete, a logic low clamp signal cl is provided to clamp switch 299 thereby uncoupling the backsides of capacitors 226, 228 from the reference voltage source Vref.
After a row of pixels has been readout, sampled, and held, then, generally in column order, the sample and hold circuits 261 output their stored signals to the amplifier 248. When reading from a first sample and hold circuit 261, a logic high control signal Φamp is provided to the feedback circuits to close switch 279 to couple the first output of amplifier 248 through capacitor 278 to its first input and to close switch 277 to couple the second output of amplifier 248 through capacitor 276 to its second input. A logic high crowbar control signal, e.g., crowbar1 for the sample and hold circuit 261 associated with the first column, is also provided to the sample and hold circuit 261 being readout to close the associated crowbar switch 239, thereby coupling the front side of capacitor 226 to the front side of capacitor 228. A logic high control signal, e.g., c1 for the sample and hold circuit 261 associated with the first column, is also provided to the sample and hold circuit 261 being readout to close switch 218 and switch 216, thereby coupling the backside of capacitor 226 to the first input of amplifier 248 and coupling the backside of capacitor 228 to the second input of amplifier 248.
After the reset and integrated charge signals have been readout to amplifier 248, a logic low control signal Φamp is provided to the feedback circuits to open switch 279 and uncouple the first output of amplifier 248 from capacitor 278 and to open switch 277 and uncouple the second output of amplifier 248 from capacitor 276. A logic low crowbar control signal (e.g., crowbar 1 for the first column) is provided to the sample and hold 261 being readout to open the associated crowbar switch 239, thereby uncoupling the front side of capacitor 226 from the front side of capacitor 228. A logic low control signal e.g., c1, is also provided to the sample and hold 261 being readout to open switch 218 and switch 216, thereby uncoupling the backside of capacitor 226 from the first input of amplifier 248 and uncoupling the backside of capacitor 228 from the second input of amplifier 248. Thus, a correlated double sampled signal is provided as output from amplifier 248 resulting from the input of the integrated charge and reset signals to the amplifier 248. After a row of sample and hold circuits 261 have been readout, a next of row of pixels 150 in the pixel array 230 are sample, held, and then readout through the amplifier 248.
FIG. 6 illustrates a modified pixel array 230′ that uses 4-way shared pixel circuitry comprising four pixels in neighboring columns and which desirably omits a row select transistor in the readout circuit for the shared pixel circuits. The pixel array 230′ is an alternative to the pixel array 230. The pixel array 230′ is comprised of even columns that include pixels 450a-d and odd columns that include pixels 451a-d. Although pixel array 230′ is depicted as including three columns and four rows, the pixel array 230′ is representative of a pixel array having any plurality of rows and columns. The columns of the pixel array 230′ are labeled Y(m+1), Y(m), and Y−1(m+1) and the rows of pixel array 230′ are labeled X(n), X(n+1), X(n+2), and X(n+3).
In array 230′ pixels are diagonally grouped by color into a pixel circuit; thus, green pixels are grouped together and blue and red pixels are grouped together. A green pixel circuit, for example PixelCircuit1, is comprised of pixels 451a, 450b, 451c, and 450d. The green pixel circuit PixelCircuit1 also includes a reset transistor 484 and a source follower transistor 486. A blue and red pixel circuit, for example PixelCircuit2, is comprised of pixels 450a, 451b, 450c, and 451d. The blue and red pixel circuit PixelCircuit2 also includes a reset transistor 485 and a source follower transistor 487. In operation, the green pixel circuit PixelCircuit1 is readout, row by row, through a single column line, e.g., Col Y(m+1) and the blue and red pixel circuit PixelCircuit2 is readout, row by row, through a single column line, e.g., Col Y(m). No row select transistors are used in the readout circuit to couple the source follower transistors 486, 487 to a column line.
Pixel array 230′ also includes transfer transistor control lines associated with each row of the array 230′, e.g., TX X(n) for pixels in row X(n) associated with transfer transistors 490a and 491a. Additionally, pixel array 230′ includes reset transistor control lines associated with each group of four rows of the array, e.g., RST X(n) for pixels in rows X(n), X(n+1), X(n+2), and X(n+3), associated with reset transistors 484, 485. Moreover, pixel array 230′ includes column pull up (Col_Pu) transistors 498 to control coupling a Vaa-pix voltage to a column line 496, 497.
FIG. 7 depicts a simplified correlated double sampling and column read out timing for the pixel array 230′ of FIG. 6. To begin a readout operation of a row X(n), at a time t1, a row address X(n) is provided to row addressing circuit 234 and column addressing circuit 244 of FIG. 2. A Col_Pu signal is applied to transistors 498 to couple lines 496, 497 to a voltage (e.g., Vaa-pix signal level) and therefore to activate the reset transistors 484, 485. At time t2, a logic high RST signal is provided to the reset line RST X(n), thereby placing a reset charge on one of a source or drain of reset transistors 484, 485. The floating diffusion regions 494, 495 are reset by this operation. At time t3, a logic low Col_Pu signal is applied to transistors 498 to turn off transistors 498 and to deactivate reset transistors 484, 485, no longer resetting diffusion regions 494, 495. Time t3 occurs approximately 250-750 ns after time t2 occurs, preferably 500 ns.
At time t4, a logic high VLN_EN control signal is provided to the gates of column line transistors 491, 492, thereby creating a pull down circuit on the associated column lines, e.g., 496, 497. Time t4 occurs 50-100 ns after time t3, preferably 70 ns. After time t4, a logic high SHR signal is strobed to sample and hold a reset signal Vrst readout of the floating diffusion regions 494, 495 into sample and hold circuitry. The SHR strobe lasts approximately 1-2 μs, preferably 1.5 μs. A logic high TX(n) then is strobed, which closes transfer transistors 491a, 490a and couples the photodiodes 462 to their associated floating diffusion regions 494, 495, thereby transferring the accumulated charge from the photodiode 462 to their associated floating diffusion regions 494, 495. The TX strobe lasts approximately 500-1000 ns, preferably 750 ns, ending at time t5. A logic high SHS signal is strobed to sample and hold accumulated charge read from the floating diffusion regions 494, 495 into sample and hold circuitry. The SHS signal begins to be strobed before the TX strobe has completed, e.g., before time t5. The strobe of the SHS signal lasts approximately 1-2 μs, preferably 1.5 μs and ends at time t6. At time t7, a logic low VLN_EN signal is provided thereby no longer creating a pulldown circuit on the associated column line. Time t7 occurs approximately 50-100 ns, preferably, 70 ns, after time t6, e.g., the completion of the SHS strobe. Subsequently, a logic high Col_Pu signal and a logic low RST(n) signal are provided. Thus, a reset signal and a charge accumulation signal are sampled from the pixel array 230′.
At t8, a rolling shutter operation occurs. A row address X(n+m) is provided to row addressing circuit 234 and column addressing circuit 244 of FIG. 2, which is used for a rolling shutter. After time t8, a logic high RST(n+m) signal and a logic high TX(n+m) are provided. The strobe of the TX(n+m) signal occurs while the RST(n+m) is provided with a logic high signal. After the rolling shutter operation ends, e.g., at time t9, the next row of the pixel array is sampled, e.g., row n+1. The pixel array continues to be readout, row by row, until substantially all of the rows of the pixel array have been readout. Thus, a reset signal and a charge accumulated signal are read out from the pixel array. Further, a rolling shutter has been toggled.
With the pixel array 230′ (FIG. 6) PixelCircuit1, PixelCircuit2, are comprised of zigzagged pixels in two neighboring columns, so the pixel circuits are asymmetric and are difficult to significantly reduce in size.
It is desirable to have a shared pixel circuit that is more compact and of reduced size.